Tupi phonology
The phonology of tupinambá has some interesting and unusual features. For instance, it does not have the lateral approximant or the multiple vibrant rhotic consonant /r/. It also has a rather small inventory of consonants and a large number of pure vowels (twelve). This led to a Portuguese pun about this language, that Brazilians não têm fé, nem lei, nem rei (have neither faith, nor law, nor king) as the words fé (faith), lei (law) and rei (king) could not be pronounced by a native Tupi speaker (they would say pé, re'i and re'i). Vowels It should be noted that the nasal vowels are fully vocalic, without any trace of a trailing or . They are pronounced with the mouth open and the palate relaxed, not blocking the air from resounding through the nostrils. These approximations, however, must be taken with caution, as no actual recording exists and it is known that Tupi had at least seven dialects. Consonants * The glottal stop is only found between a sequence of two consecutive vowels and at the beginning of vowel-initial words (aba, y'', ''ara, etc.). When indicated in writing, it is generally written as an apostrophe. *† Some authors remark that the actual pronunciation of was retroflex . Additionally, most sources describe some dialects having /s/ and /h/ in free variation. *‡ The actual pronunciation of ŷ is the corresponding semivowel for . It appears to be unique to Tupi–Guarani languages, and might not have existed in all dialects. An alternative view According to Nataniel Santos Gomes, however, the phonetic inventory of Tupi was simpler: * Consonants: ** p, t, k, ‘ ( ) ** b ( ) ** s, x ( ) ** m, n, ñ ( ) ** û (/w/), î (/j/) ** r ( ) * Vowels ** i, y ( ), u, ĩ, ỹ, ũ ** e, o, õ, ẽ ** a, ã This scheme does not regard Ŷ as a separate semivowel, does not consider the existence of G ( ), and does not differentiate between the two types of NG ( and ), probably because it does not regard MB ( ), ND ( ) and NG ( ) as independent phonemes, but mere combinations of P, T, and K with nasalization. Santos Gomes also remarks that the stop consonants shifted easily to nasal stops, which is attested by the fitful spelling of words like umbu (umu, ubu, umbu, upu, umpu) in the works of the early missionaries and by the surviving dialects. According to most sources, Tupi semivowels were more consonantal than their IPA counterparts. The Î, for instance, was rather fricative, thus resembling a very slight , while Û had a distinct similarity with the voiced stop , thus being sometimes written gu. As a consequence of this character, Tupi loanwords in Brazilian Portuguese often have j'' for Î and ''gu for Û. Considerations on the writing system It would have been almost impossible to reconstruct the phonology of Tupi if it did not have a wide geographic distribution. The surviving Amazonian Nhengatu and the close Guarani correlates (Mbyá, Nhandéva, Kaiowá and Paraguayan Guarani) provide material that linguistic research can make use of today in order to achieve an approximate account of the language. Scientific reconstruction of Tupi suggests that Anchieta simplified (or merely overlooked) the phonetics of the actual language when devising his grammar and his dictionary. The writing system employed by Anchieta is still the basis for most modern scholars. It is easily typed with regular Portuguese or French typewriters and computer keyboards (but not with character sets such as ISO-8859-1, which cannot produce ẽ'', ''ĩ, ũ'', and ''ỹ). Its key features are: * The tilde indicating nasalisation: a'' → ''ã. * The circumflex accent indicating a semivowel: i'' → ''î. * The acute accent indicating the stressed syllable: abá. * The use of the letter x'' for the unvoiced palatal fricative , a spelling convention common in the languages of the Iberian Peninsula, but unusual elsewhere. * The use of the digraphs ''yg (for Ŷ), gu (for ), ss (to make intervocalic S unvoiced), and of j to represent the semivowel . * Hyphens are not used to separate the components of a compound, except in the dictionary or for didactical purposes.